Fine polar invariants of minimal singularities of surface Romain Bondil To cite this version: Romain Bondil. Fine polar invariants of minimal singularities of surface. 2004. hal-00001089 HAL Id: hal-00001089 https://hal.archives-ouvertes.fr/hal-00001089 Preprint submitted on 30 Jan 2004 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. FINE POLAR INVARIANTS OF MINIMAL SINGULARITIES OF SURFACES ROMAIN BONDIL Abstract. We consider the polar curves PS,0 arising from generic projections of a germ (S, 0) of complex surface singularity onto C2. Taking (S, 0) to be a minimal singularity of normal surface (i.e. a rational singularity with re- duced tangent cone), we give the δ-invariant of these polar curves, as well as the equisingularity-type of their generic plane projections, which are also the discriminants of generic projections of (S, 0). These two (equisingularity)-data for PS,0 are described in term, on the one side of the geometry of the tangent cone of (S, 0) and on the other side of the limit-trees introduced by T. de Jong and D. van Straten for the deformation theory of these minimal singularities. These trees give a combinatorial device for the description of the polar curve which makes it much clearer than in our previous Note on the subject. This previous work mainly relied on a result of M. Spivakovsky. Here we give a geometrical proof via deformations (on the tangent cone, and what we call Scott deformations) and blow-ups, although we need Spivakovsky’s result at some point, extracting some other consequences of it along the way. Introduction The local polar varieties of any germ (X, 0) of reduced complex analytic space were introduced by LˆeD.T. and B. Teissier in [17]. In particular, the multiplicities of the general polar varieties are important analytic invariants of the germ (X, 0). However, as also emphasised by these authors (see also [23] p. 430–431 and [24]), there is more geometrical information to be gained on the geometry of (X, 0) by considering not only the multiplicity but the (e.g. Whitney)-equisingularity class of these general polar varieties, which can be also shown to be an analytic invariant. In this work, we will focus on the polar curves of a two-dimensional germ (S, 0). Our reference on equisingularity theory for space curves will be the m´emoire [8]. Of course, as opposed to the case of plane curves, there is no complete set of invariants attached to a germ of space curve describing its equisingularity class. As a general rule, results on equisingularity beyond the case of plane curves only make sense by considering the constancy of invariants in given families. Here we look at the family of polar curves and will consider the following invariants : Definition 0.1. Our equisingularity data for a germ of space curve is both : (i) the value of the delta invariant of the curve, (ii) the equisingularity class of its generic plane projection. ccsd-00001089 (version 1) : 30 Jan 2004 2000 Mathematics Subject Classification. Primary : 32S15, 32S25, Secondary : 14H20, 14B07. Key words and phrases. rational surface singularity, minimal singularity, polar curve, discrim- inant, limit tree, deformation, tangent cone, Scott deformation. 1 2 ROMAIN BONDIL We recall the definitions of these notions in the text (see def. 6.1 and def. 1.2). The constancy of these two invariants in a family of space curves ensures Whitney conditions and actually the stronger equisaturation condition (cf. [8]). In general, this is still partial information : for example, another interesting invariant for space curves, namely the semi-group of each branch, is completely independent of this equisaturation condition. The purpose of this paper is to describe the equisingularity data in 0.1 for the the general polar curve of a class of normal surface singularities called minimal. These minimal singularities were studied in any dimension by J. Koll´ar in [19]. In the normal surface cases, they are also the rational singularities with reduced fundamental cycle and were studied by M. Spivakovsky in [21] and T. de Jong and D. van Straten in [15]. For these surfaces, we prove that the general polar curve is a a union of An-plane curves singularities1, where the n’s and the contacts between these curves can be deduced from the resolution graph of the surface (∗). This information gives in particular a complete description of part (ii) of the data in 0.1, i.e of the general plane projection of the polar curve, which is also the discriminant of the general projection (the coincidence of these two concepts is a theorem, cf. section. 1). The information on the discriminant was already given in the Note [5] as a consequence of a result of Spivakovsky, but the statement there was clumsy. Here we give a much nicer device that allows to read directly the information about of the discriminant (or the polar curve as well) from both the information contained in the tangent cone of these singularities and the one given by a graph deduced from the resolution graph, which is precisely the limit tree introduced by T. de Jong and D. van Straten in their study of the deformation theory for these minimal singularities (see [15]). We also provide an inductive proof relying much more on the geometrical prop- erties of these minimal singularities. This proof makes the core of the paper. It still uses Spivakovsky’s theorem however, mainly through a characterisation of generic polar curve on the resolution which we deduce along the way. The several plane branches of the polar curve lie in distinct planes in a bigger linear space, and the value of the delta invariant (part (i) in 0.1) gives some (partial) information on the configuration of these planes in the space. We explain how this delta invariant is easily computed from the what we call the Scott deformation of the surface, which turns out to give a delta-constant deformation of the polar curve onto bunches of generic configurations of lines. Organisation of the paper : In section 1, we recall the definition of the general polar curve PS,0 of a germ of 2 surface (S, 0), of the discriminant ∆S,0 of a generic projection of (S, 0) onto C and the important result that ∆S,0 is a generic projection of the curve PS,0. Section 2 gives the definition of minimal singularities in general, the particular case of normal surfaces, and their characterization by their dual resolution graph. We then define in section 3 a notion of height on the vertices of this resolution graph, which was used in other places such as [21] and [15], and corresponds to the number of point blow-ups necessary to let the corresponding exceptional component appear. We also give there our convention in representing dual graphs with • and 1hence the information about the semi-group if the branches is obvious FINE POLAR INVARIANTS OF MINIMAL SINGULARITIES OF SURFACES 3 ∗ and define reduced dual graphs to be the ones in which the self-intersection for components of the tangent cone has the minimum absolute value. In section 4, we give the description of generic polar curves on a resolution of a minimal singularity as proved by M. Spivakovsky (thm. 4.2). This result will play the following somehow different roles in the sections following it : (i) Section 5 explains how, using the full strength of this result, one may derive quite quickly a description of the generic discriminant ∆S,0 (more precisely of the statement (*) for the polar curve). This sums-up the Note [5] in an improved way, and an mistake in an example there is corrected. (ii) In section 6, we mention how, using a result of Giraud, theorem 4.2 also permits, at least theoretically, to deduce the δ-invariant for the general polar curve from the shape of its transform on the minimal resolution. This result is however not useful for concrete computations, for which we use another approach in section 11. (iii) In section 7, we get, as a purely qualitative consequence of (i) and (ii), a characterisation of generic polar curves on a resolution XN of both the singularity (S, 0) and its Nash-blow-up. This will be the application of Spivakovsky’s result we will use in the proof of our main result. Section 8 to 11 form the core of the text : • in section 8, the polar curve for the tangent cone of a minimal singularity is made geometrically explicit and through the process of deformation onto the tangent cone is also seen as part of the polar curve of the singularity. • in section 9, we recall what we need from the limit tree construction of de Jong and van Straten. With this, • in section 10, we give, and prove, our main theorem describing the information in (∗) using the limit tree construction and the contribution of the tangent cone. • in section 11, we show how a special deformations of minimal singularities has a nice interpretation in our descriptions of polar curves and also give a nice way to compute the delta invariant of these, completing the information in def.